Using the ODP Bootstrap Model: A Practitioner’s Guide

There are many papers that describe the over-dispersed Poisson (ODP) bootstrap model, but these papers are either limited to the basic calculations of the model or focus on the theoretical aspects of the model and always implicitly assume that the ODP bootstrap model is perfectly suited to the data being analyzed. In order to use the ODP bootstrap model on real data, the analyst must first test and review the assumptions of the model and may need to consider various modifications to the basic algorithm in order to put the ODP bootstrap model to practical use. This monograph starts by gathering the evolutionary changes from different papers into a complete ODP bootstrap modeling framework using a standard notation. Then it generalizes the basic model into a more flexible framework. Next it describes the adjustments or enhancements required for practical use and addresses the diagnostic testing of the model assumptions. While this monograph is focused on the ODP bootstrap model, we must recognize that it is a special subset of a larger framework of models and that there are a wide variety of other stochastic models that should also be considered. However, since no single model is perfect we also explore ways to combine or credibility weight the ODP bootstrap model results with various other models in order to arrive at a “best estimate” of the distribution, similar to how a deterministic best estimate is generally derived in practice. Finally, the monograph will also extend the model to illustrate the GLM Bootstrap and the model output to address other risk management issues and suggest areas for future research.

Keywords: Bootstrap, Over-Dispersed Poisson, Reserve Variability, Reserve Range, Distribution of Possible Outcomes, Generalized Linear Model, Best Estimate.